Tiny Core Linux
Tiny Core Base => TCB Talk => Topic started by: philip on March 28, 2009, 12:23:59 PM
-
TC is a nice platform for users like me to learn how GNU/Linux systems work. I am not quite ready for Linux From Scratch, but I would love a chance to take a step closer, starting with TC.
Is it accurate to think about the TinyCore Base as a collection of truly minimal system ingredients (kernel, libc, busybox, wget, robertscripts, etc) plus a collection of carefully-selected extensions (tinyx, jwm, etc) that most users will want? If so, what's the smallest setup that can boot and start loading extensions, presumably driven by command-line instructions? Would it be feasible to turn that setup into an explicit packaging option?
In my dreams, I download and boot a microcore.iso that gets me online, and then start adding extensions that could take me up to a standard TCB setup or, perhaps, in some other direction. This way of doing things would always be just an option for experienced users to play with, and most definitely not the first point of contact for newcomers. The current choices in the Base are excellent for most purposes, and the nearly-instant appearance of a friendly GUI is a huge attraction of TC as it now stands.
This idea may be worth the developers' time:
- If it's close to the way they build TC now anyway, sharing a peek behind the scenes wouldn't take much extra work.
- It may recruit to the community curious and technically adept people who then contribute to the general advancement of the project.
Please discuss. Thanks.
-
TC is a nice platform for users like me to learn how GNU/Linux systems work. I am not quite ready for Linux From Scratch, but I would love a chance to take a step closer, starting with TC.
it's not just nice for learning, it really gives you a feeling of having choices again. don't uninstall that application or library, either don't install it, or just delete the file (or move it to optional.) i like your idea, although it's not the most important thing for tc i believe. the most important thing right now is that it become useful for a wider variety of people and have more applications (choices.) i think the idea of making it even more modular is good for 2.0 or later, you could still have the very basic "desktop" in the main iso, instead of bzImage and tinycore.gz, have bzImage, tinycore.gz, and tinycore.tgz (for the basic desktop) all in the iso (and the frugal install.) to make it micro, remove tinycore.tgz. i think including the desktop for most users (less obsessed than we) is better.
-
My ideal MicroTinyCore would be have no desktop or client apps other than enough for an xserver and ssh client capable of handling x11 forwarding from the fully loaded MonsterTinycore that's serving up all my client apps... Wahoo!
-
This too has been much discussed. But initially I wanted to appeal to a wider audience.
I wanted to offer a somewhat familiar desktop onto which applications could easily be integrated at boot and upon demand.
If I had started from the CLI, no X, I could just hear the reviews, a 1970 version of unix, too difficult for any but the most advanced users. However, I realize there are some, perhaps many that pushing my core concepts to even a lower level would be most appealing.
We will have to see. At the moment, there is still much to accomplish with the current offering.
-
TC is a nice platform for users like me to learn how GNU/Linux systems work. I am not quite ready for Linux From Scratch, but I would love a chance to take a step closer, starting with TC.
I'm about half way through lfs now, and it's fairly easy. I resisted bothering with it until now, but if you really want something simple where you are in control of updating the core libs when necessary, I think it's a good idea. I think tc needs to be more sophisticated, as Robert indicated.
-
I think tc needs to be more sophisticated, as Robert indicated.
I agree with both of you, and my initial post says the same thing. It's interesting, though, how the term "sophisticated" could spin both ways. The standard public version of TC should clearly have the full functionality it offers now: that makes it sophisticated in its operation. The barebones version I am suggesting as an add-on (not a replacement!) would require sophistication from its users, rather than in the ISO.
Thanks to all who have contributed so far. Let's keep this interesting topic alive.
-
It becomes a silly game, really. TC loads 40 loops almost instant on. You can remaster TC in about 5 minutes, and strip out everything you don't want. Don't forget, it takes some time to boot your computer up to the grub menu.
I think a few people get overly hung up on "tiny", to the detriment of what an OS is for. I suppose it's possible to have an 8M core that doesn't work with anything. It'll boot up .5 seconds faster than TC does now. Look at all the work that went into keeping DSL small, and check out their latest ratings. IMHO, the core should have been upgraded a long time ago. TC works with modern software/hardware, and will rapidly diminish in value to the extent that it only supports "old" applications/hardware....even if it's only 10M.
-
Sizewise it really doesn't matter whether we are 10mb or 8mb without X, jwm, fltk.. Everybody has at least 16 mb CF cards ;)
Even for burning into a bios chip, for which TC is not quite meant for, wouldn't work, since most bios chips are 1mb.
As for the public iso, I think it would be rather easy to keep both additional tce directory support and to have jwm.tcz et al loaded to ram inside tinycore.gz in order to still stay boot device-agnostic. Just, say, putting them to /tmp, and having some tce-loads in bootlocal.sh. With our current tcz loading speed it wouldn't hurt much at all.
@jpeters: why so negative prediction? updates will come as necessary (and no, flash is not necessary), in fact soon the kernel .29 might be seen around here. More and better HW support, I'd say ;)
-
Just for kicks I removed out of tinycore.gz all that relates to X, the fltk apps, and brought the size of the iso down to 6.3MB. I put all that I removed into an extension so I could have it back with one command. I booted in 10 seconds on my modest machine and was sitting on 18.7MB of used RAM. I added the extension back on and then started X and added my usual apps like nothing was different.
In theory such a setup would boot on a 24MB machine and could use tce's/tcz's for server functions like nfs-utils or other small apps, theory because I don't own a machine that size but I am sure it would work. A simple script, like restore-desktop or other such name, could fetch and restore what was missing. Or one could hand pick what they wanted if they knew what they were doing, but if you bring in any of the X related stuff it all kind of goes together. One would not have to very savvy to boot with a text mode TC, but one would have to be a little bit to make use of a text mode TC outside of restoring the desktop.
While I am not pushing for this to happen officially, a text mode tinycore makes perfect sense to me and is not hard to make for oneself. I looked to see what in base depends on libstdc++ besides the fltk apps and didn't see anything but I could have overlooked a dependency on it. That would bring the iso size to a little over 6.5Mb if it was put back in.
-
@jpeters: why so negative prediction? updates will come as necessary (and no, flash is not necessary), in fact soon the kernel .29 might be seen around here. More and better HW support, I'd say ;)
I think it sounded more negative than intended. I began thinking (yes...after becoming aware of the glibc issue..) that it might be nice to have two versions going, one focused on 'small as possible' and another focused more on Robert's brilliant modular approach, yet maintaining updated core libraries (eg, Glibc, GCC, Binutils) and hardware support. I'd predict the latter will make it up the charts in DistroWatch (if anyone cares).
Meanwhile, I'm learning how to make my own. (BTW....I'd highly recommend recommend Robert Love's book, "Linux Kernel Development", which is actually readable for hobby-ists like myself).
-
... and brought the size of the iso down to 6.3MB. I put all that I removed into an extension so I could have it back with one command.
... I added the extension back on and then started X and added my usual apps like nothing was different.
I just started to play with TinyCore, so I don't have enough knowledge to participate in the current topic,... but...
I like the idea Jason is using...
Maybe, in the future, for developers alike, it is possible to have some kind of "TinyCore Compiler/Builder" that facilitate the user to "choose" versions and components always using the same infrastructure...
Something like this:
(http://farm4.static.flickr.com/3197/3397710998_b0bd388e44_m.jpg)
[higres]
http://farm4.static.flickr.com/3197/3397710998_48cce93375_o.jpg
It maybe possible to have a "torrent repository on-demand"...
Or, some kind of TinyCore LiveCD Builder, that allows to "build" the kind of TinyCore we want...
Anyway,... just some ideas to consider for the long run...
-
i don't think tinycore needs to be more sophisticated, i think almost everyone forgets what it's like to be a noob, or, they were a noob when the word meant a lot more than it does now.
these days, a noob may boot ubuntu and simply wonder how to uninstall applications, or why the term is useful. he's not a noob that is going to start compiling x11 or installing drivers, he's not going to think about the command line for tasks that a gui will do just as well as windows. he's not going to try to customize every aspect of the system if he can learn to use the defaults. depending on when you first used linux, being a "noob" means you know much less than you would if you were new to linux in the earlier days- it also means you can do more, easily.
tinycore gives you a pretty friendly environment, some people will think "oh great, it looks like that old windows machine from the 80's" when they see jwm, they've never even used an alternative shell in windows.
tinycore shows people they can have something different, but not totally unfamiliar. i think it's very good to say it's a desktop distro, not a distro that allows you to install a desktop if you want. but the way i suggested, it would let you get to a microcore by deleting one file, (if you did a usb install) or remove one file from the iso, or, it would be very easy to have two isos. but i don't think the first thing *most* people should see, when they live in a world of huge graphical distros- is the command line unless they want to.
tinycore has a nice looking graphical installer for everything, appbrowser. someday it will be able to install nicer looking window managers (although jwm is my favorite.) appbrowser should be the standard experience for the average guy trying tc for the first time. you won't get it from the command line- and really, you shouldn't. but it should be very, very easy to get to microcore from the "average," standard 10mb version, or there should be a microcore iso too.
-
Or, some kind of TinyCore LiveCD Builder, that allows to "build" the kind of TinyCore we want...
That would be Robert......
-
That would be Robert......
actually the mkmydsl article seems to allude to some online custom dsl builder. i've never seen one though, or read about it anywhere else. in the future it should be easier to create websites that do custom distro builds before you even download the .iso. in tc this would mean it would be easier to have a few extra drivers at boottime.
-
MydlsMaker? I never saw it before either. As the article noted however, it's much easier to make the iso locally, just altering your own setup. How long does it take to do a tc remaster?
-
The GUI that netzen described would be pretty easily doable. Rather hard in windows, of course, as that OS is not able to loop-mount images natively, but for them, there could just be a zip with the current contents. And a binary of mkisofs.
Essentially, it would be a GUI to unpack the iso, then a frontend / caller for appbrowser to download all desired extensions to a directory named "tce" in the root of the to-be-iso, and then a caller to mkisofs with a progress bar.
Or, it would only present a graphical form of some rather easy command line tasks.
While I'm not sure I'd like to see all kinds of Windows users then coming to ask what an ISO image is, I'm not completely against such a gui either. In my own opinion it's not important enough for me to do, but if someone does, I'll support them.
-
How long does it take to do a tc remaster?
depends on how much one has to learn. tinycore is based on the idea of: build it your own way without remastering (mydsl was too.) the best argument for any core apps is that most people do not want so much customization that they have to organize what goes into the absolute base install. (plus you need something like wget.) it takes more time and knowledge to remove a web browser and remaster than to add it- it's might be worth making remastering as easy as adding applications, where possible.
i don't think someone should ever need to remaster (or go to the trouble of setting up persistence) before having a friendly gui, i would prefer that all the gui stuff be shoved into an extension, and the extension be included in the iso, outside tinycore.gz
then you could have a bootcode that didn't use the extension... remembering that tc doesn't load the entire iso into memory, just tinycore.gz and any extensions that it finds. when someone new is setting up tc for the first time, i want them to have a gui, regardless of whether they're online yet or not. you're almost certainly thinking of what it's like to use tc after you're familiar, after you're online. i'm thinking of what it's like for someone who's never seen linux before. i want tc to continue make a first impression that isn't text-only. but if the gui was an extension, making two isos for download would be trivial.
-
then you could have a bootcode that didn't use the extension...
Yes. How about 'Secure GUI Terminal'. Tiny Core might one day be both the most minuscule and maximal distro.
It would also up the ante for ultimate bragging rights and benefit antique PC owners. Then again, they are already giddy.
The most practical application (my fantasy) will be on my NanoNetBook; a bootable wrist watch, that wirelessly connects to: the internet via satellite, my sun-glasses (the monitor!) and the paper-thin touchpad/keyboard which is unrolled from its pencil-sized container and placed on my beach-side drink table.
Don't laugh. It will happen.
-
Why is that alcohol never goes out of fashion?
-
Ha. Well, that was MY fantasy.
Regarding Robert's reality check comments, what change do you think would give the most bang-for-the-buck to increasing general user popularity?
-
The most practical application (my fantasy) will be on my NanoNetBook; a bootable wrist watch, that wirelessly connects to: the internet via satellite, my sun-glasses (the monitor!) and the paper-thin touchpad/keyboard which is unrolled from its pencil-sized container and placed on my beach-side drink table.
Don't laugh. It will happen.
the technology that replaces the technology that replaces bluetooth will probably have enough bandwidth for your sunglasses. japan already has monitors like that for consumer use, what i would like to see is more open standards for electronic interfaces (not just for the connection but also the device) so that open drivers are easier to write.
otherwise it will be proprietary hardware and second-rate drivers, whether buggy closed-source drivers that work, or open source drivers that work but have fewer features. i like the free software movement, it could be helped by an open hardware movement. open hardware projects (bios, car, phones (http://www.opencellphone.org/index.php?title=Main_Page)) but there is no organization to bring them together.
with the passionate nature of philosophers like richard stallman, and the practical, sometimes compromising nature of people like bruce perens and lawrence lessig, some would argue it's better that there is no free hardware foundation. i understand the sentiment. it would be nice to see open hardware conventions (meetings) though. i'd love to build an open source device to run tinycore on, something that looked like a miniature commodore 64 i could put in a carry case and connect to a variety of monitors, including a very small one for more portable use, and a standard monitor for home.
-
Just for kicks I removed out of tinycore.gz all that relates to X, the fltk apps, and brought the size of the iso down to 6.3MB. I put all that I removed into an extension so I could have it back with one command. I booted in 10 seconds on my modest machine and was sitting on 18.7MB of used RAM. I added the extension back on and then started X and added my usual apps like nothing was different.
Could you please upload the 6.3MB version? It would be great.
Still, I'm not an expert but hopefully this will help me with my project.
Or maybe a quich how to... I'm still new to most of the stuff (especially to strip downs like this)
Thanks & TinyCore is great! Exactly what I searched. I already spent some hours remastering it... :)
-
I think I deleted the image as well as the list as it was mainly an experiment, but I will look once more. I would rather not post an iso or tinycore.gz but if I see the list I will upload it. But I think it is gone.
-
The most practical application (my fantasy) will be on my NanoNetBook; a bootable wrist watch, that wirelessly connects to: the internet via satellite, my sun-glasses (the monitor!) and the paper-thin touchpad/keyboard which is unrolled from its pencil-sized container and placed on my beach-side drink table.
Don't laugh. It will happen.
Do away with the wristwatch - make the keyboard bootable & have the wireless connectivity. Also make it waterproof, so you can go swimming with it. And solar powered.
-
A microcore option comming soon; currently in alpha testing. Stay tuned...
-
great news. can't wait :)
-
Do away with the wristwatch - make the keyboard bootable & have the wireless connectivity. Also make it waterproof, so you can go swimming with it. And solar powered.
You want to go swimming with your computer? I do like specific devices that I can place on my surfboard to track speed, etc, but don't think I care to read my email while swimming. It's always interesting to check out ideas that have failed...usually because they really don't add much to peoples' lives. I've been waiting for a good ebook reader for years....We've got Sony and Kindle, but still lousy graphics.....go figure..but then videos, twitter, and itunes have more popular appeal then reading.
-
Do away with the wristwatch - make the keyboard bootable & have the wireless connectivity. Also make it waterproof, so you can go swimming with it. And solar powered.
You want to go swimming with your computer? I do like specific devices that I can place on my surfboard to track speed, etc, but don't think I care to read my email while swimming. It's always interesting to check out ideas that have failed...usually because they really don't add much to peoples' lives. I've been waiting for a good ebook reader for years....We've got Sony and Kindle, but still lousy graphics.....go figure..but then videos, twitter, and itunes have more popular appeal then reading.
I've seen youtube videos of people waterproofing their laptops to use by the pool when dripping wet. And in the bath. I personally bought a waterproof keyboard after my kids spilled juice over one. Anyway - with global warming and sea level rises, it might be more important than we think!
-
Shaking hands and increased coffee addiction are good reasons for waterproof keyboards too.
-
Greetings.
I agree with jpeters. It should not be a size game. It is about best core system that can be expanded in a modular way to suit user's needs.
I am a new user of TCL and after playing a lot over years with Wolvix, Vector, Zenwalk, I discovered Austrumi and then Slitaz. As bloat is not acceptable to me I thought I found the best Linux until I dicovered TCL a few weeks ago. I am not about looking for something to run on extremely old hardware but on any hardware, old and newest, a system with dexterity, modularity and integrity. From my testing TCL is absolutely the best, both in concept and implementation.
Some may not understand or notice how well TCL is designed and implemented. Let me just mention one example out of many, namely network support:
I put TCL in my desktop - recognized and configured my gigabit network. I put it in my old thinkpad 600 with PCMCIA network card - recognized and configured! Connected to the Internet without ado! Neither Slitaz nor Austrumi nor Zenwalk could do this! And I am mentioning just the best ones!
Modularity: Outstanding. You can put together anything you need and only what you need in a short time and easy...
Integrity: The system you put together can be easily maintained in a pristine state, preventing corruption.
The creator of this gem really knows what is the right thing to do! I just hope this great work will continue and will be available to larger audience.
I am concerned that considering the TCL creator's health this will not be abandoned in the future...
-
the idea is good, and i don't think that this would be just a size game, but a way to have even more modularity. i am waiting for the option which Robert speaks about.
-
Shaking hands and increased coffee addiction are good reasons for waterproof keyboards too.
..also tears..
-
Quote from: curaga on Today at 06:03:28 AM
Shaking hands and increased coffee addiction are good reasons for waterproof keyboards too.
..also tears..
..and also beers.. But I can get old hardware out of the recycle box at work so no real loss here.